Instant Voice Clones is now live on Voice-Swap. All you’ve got to do to create your own is upload or record ten to twenty seconds of clean speech, and your voice is ready to use in our professional workspace, Voiceover_Studio.
Clones appear in the voice selector alongside our built-in library, and you direct them the same way you direct any voice in Studio: multiple takes, pauses where you want them, and reactions like laughs and sighs dropped into the read.
Create your first Instant Voice Clone here.
What you need
Ten to twenty seconds of speech — ten seconds of actual talking is the minimum. Silence at either end is trimmed automatically and doesn't count towards it, so combine a few short clips if one on its own falls short.
Uploaded or recorded — drag and drop .wav, .mp3, .m4a, .aiff or .flac, or record straight into the browser.
A corrected transcript — we transcribe the clip for you, and you review it before the clone is created. Word-for-word accuracy matters more than punctuation; the transcript is what the voice is aligned to, so a misheard name or a dropped phrase costs you quality.
Why we built it
Our high-end voice models remain the heart of the platform, and for professional work, they are still our gold standard. Trained on a high-quality dataset with the talent’s full participation, they hold up across long-form reads, audiobooks, podcasts and more, while carrying the fidelity that professional delivery demands.
Instant Voice Clones are slightly different. They are very usable for quick referencing, demoing, testing and editing but they might not have the flexibility or fidelity of our flagship models. The two are complementary. An Instant Voice Clone is quick, personal and usable. A trained model is an asset with rights and potential revenue attached. It’s up to you how you want to work with Voice-Swap. Contact us here if you’re interested in building your own HQ voice model.
What makes a good model
The clone mirrors what it hears, so read at the pace and energy you want back, and record with the quality you want the model to portray. Record calmly, and you get a calm read. Record upbeat, and you get upbeat. Your audio clip is a performance direction as much as a voice sample – treat those twenty seconds like a take, not a file upload.
To create an Instant Voice Clone, you need to be subscribed to our Advanced tier and above, which you can do here. Once you’re done, create your own Instant Voice Clone here.
Need assistance, contact us here.